基于视觉显著性与霍夫森林的腰椎间盘定位与分割

Lumbar Disc Localization and Segmentation Based on Visual Saliency and Hough Forest

为了对腰椎疾病的计算机辅助诊断提供腰椎间盘的具体位置和形状信息,提出一种基于CT图像的腰椎间盘定位与分割算法.首先针对腰椎CT图像中骨骼干扰较多、背景纹理复杂的特点,使用小波分解获得局部特征,并引入独立元分析方法将高维特征空间分解为相互独立的多个一维特征空间,提出基于视觉显著性检测的腰椎间盘区域预选取算法,实现了小波特征的概率密度估计,并生成了腰椎间盘显著性分布图;然后提出基于显著性分布与方向信息测度相融合的局部样本随机提取算法,同时设计了面向稀疏特征的弱分类器来提高随机森林的分类准确性,从而实现腰椎间盘区域的精确定位与分割.实验结果证明,该算法能够有效地提高腰椎间盘检测的准确率与计算速度.

In order to provide the specific information of location and shape of lumbar discs for the computer-aided diagnosis of lumbar disease, an algorithm of lumbar discs localization and segmentation based on CT images is proposed in this paper. In consideration of the disturbance of skeleton and complicated textures of the background, local features are extracted using wavelet decomposition, and the high-dimensional feature space is transformed into assemble of multiple one-dimensional feature spaces by introducing independent component analysis, a pre-selection algorithm for lumbar discs based on visual saliency detection is presented to realize the evaluation for the possibility density of the wavelet features and the saliency map of lumbar discs is obtained. Besides, the local samples are randomly extracted based on both the saliency map and the orientation information measurement, and a new category of weak classifiers for sparse features is designed to improve the recognition rate of the Hough forest algorithm and to realize the accurate localization and segmentation of lumbar discs. The experimental results show that the proposed algorithm is effective in improving the accuracy and the computational speed of lumbar discs localization and segmentation.